DFI LanParty NF4 Ultra-D Problems

Cyclonite

Hey there.

I'm having an issue that seems to have slowly manifest itself over the past couple months. It finally hit full force this weekend.

At first, anytime I made a hardware configuration change; specifically, adding another hard drive or two, it seemed a little iffy whether it was stable or not. I would have to make BIOS adjustments or reset the CMOS, and then it would finally work.

Once I was in Windows, I never really had any issues, and I never restart my computer much. On a few occasions, when I did restart, it would require me to reset it once or twice before it booted "properly."

Now, it just finally bit it Fiday before I left for work. I restarted the computer and it just froze at the POST screen showing the CPU speed and type. I was gone until last night (Saturday), where it would not even show anything on the screen anymore. This morning, I took the board out of the case and went back to basics: Motherboard, processor, one stick of memory, and the video card. I finally got it to POST, but it would go to a black screen if I tried to get in the BIOS or past the POST screen to boot anything. After a CMOS reset or two, I finally got into the BIOS.

I wanted to flash to the latest revision, so I made a boot CD, but as soon as I plugged the optical drive in, it started to just black screen again when trying to enter the BIOS or boot off anything. Now, it actually stops sending a video signal, instead of black-screening.

I don't think it's the memory. My three suspicions are the motherboard, power supply, or processor. I'm not so sure about the processor, though. Unless, it's the memory controller.

I'm going to go to work to grab another power supply. I hope it's the problem. I've been using an Austin 450W PSU that came with my Chieftec case for a couple years now, and it's never had any problems. I know a few people are probably going to yell at me for that one, but it's worked beautifully, and I didn't do a whole lot of OC'ing.

Anyway, if it IS the PSU, I'm looking at the OCZ PowerStream 520W as a replacement. I'd like suggestions there too, but that's a side point.

Any suggestions for tests to try, I'm open and willing. I don't have a place to test my memory right now, but I'll try sometime this week.


Specs:

DFI LanParty NF4 Ultra-D Motherboard
AMD Athlon46 3200+ Winchester Core
1 GB (2x512) OCZ DDR 466 Kit (TCCD)
XFX Geforce 6600GT PCI Express

Everything at stock speeds. I was not overclocked at all.

If you need anymore information, please let me know. Thanks for your help, everyone.

//Edit: Okay, it looks like the PSU is fine. I reseated the CPU and was able to flash the BIOS, but now it's back to losing video signal after POST, no matter what. So, the problem... CPU or motherboard? I suppose I can RMA the CPU. I'm happy I went retail.

Cyclonite

Ugh. Now I have no idea. I disconnected the PS/2 keyboard and reset the board. It now seems to be fine again. I'm going to try reconnecting my primary HDD and see what happens. This is so confusing. *Sigh*

//Edit: Didn't make it. At first, I got an "error loading operating system." So, I reset, and it went black right before the XP splash. Now, it's back to black-screening after POST.

So, is that more common when the motherboard is an issue, or the processor

lemonlime

Hi Cyclonite,

The OCZ PowerStream 520 is a fantastic PSU, and I'd highly recommend it. Could you provide some more details on your Austin 450W? I.e is it a 20pin or 24 pin PSU. +12V amperage rating etc? Also, are you using all four power connectors on the board? Using the floppy and standard molex plug can greatly improve system stability.

Omega65

Cyclonite wrote:

So, is that more common when the motherboard is an issue, or the processor

The Motherboard.....but it may be a dying BIOS battery

How many Amps on that 12v rail for that PSU?

profdlp

It may be a dead end, but I'd really recommend a couple passes of Memtest-86 just to rule out a RAM problem.

My first guess would be a failing PSU, though.

Cyclonite

Wow. I never paid a lot of attention to the PSU. You think I would have. I'm only looking at 18A on the 12V rail. It's a 20-pin.

I am, and have been, using all power connectors on the motherboard.

I will grab a new CMOS battery tomorrow when I'm at work.

muddocktor

I think it's a power issue myself, because until yesterday I was scratching my head fighting no boot problems with my Dual Sata 2 system, which has an X2 4400 in it. The box I'm using it in has a Sparkle (Fortron Source) 460 psu, but the 12v rail doesn't have a very high rating on it, about like yours. It was running fine when booting at around 2400 MHz and 1.4v vcore, but after I did a volt mod to increase max vcore to 1.55 volts, it would try to power up for a second then shut down. So I kept removing the mobo from the case and testing on a box with just the vid card and ram (and a loose psu I have laying around) and it would boot and get to the bios screen, no problem. I even didn't have any problems with it booting while using a ddr booster and the ram being fed with 3.4v and overclocking it to 230 HTT. So I put it back in the case and then no boot. Finally I figured out that the Sparkle psu's 12v rail was just to wimpy to handle theextra inrush current when powering up with the proc overclocked more and with more vcore. I'm having no problems with it running on that old Enlight 350 psu at 2530 MHz and I've been doing some heatsink testing with it since figuring this out.

The Sparkle actually isn't a bad psu, but rather one that has dual 12v rails. The rail for the cpu is just too wimpy to feed the AMD dual core beast. After I finish testing heatsinks, I guess I'll have to play musical psu's and switch it out to another rig and put a psu with a beefier 12v rail in the case.

Mt_Goat

Austin / Foxcon PSU's are CRAP!!!!!!!!!! I have had horrible problems every time I have ever tried them as they were common in a lot of cases I bought. I just trash them and put something goon in its place. The OCZ you mentioned is good a are lots of others. I firmly beleive this is a good place to start.

Cyclonite

Ah! Thanks, mtgoat. I had never heard of the brand, and I figured it was a crappy PSU. I've been looking to upgrade for a while, but always put it elsewhere, because, for all intents and purposes, the PSU was running everything just fine. At least, I thought so. I guess I'll have to hunt down a more powerful PSU. I really like the OCZ a lot, but I don't really want to drop $130. *Sigh* I think I'm going to have to wait until next paycheck. Hehe.

Omega65

muddocktor wrote:

The Sparkle actually isn't a bad psu, but rather one that has dual 12v rails. The rail for the cpu is just too wimpy to feed the AMD dual core beast. After I finish testing heatsinks, I guess I'll have to play musical psu's and switch it out to another rig and put a psu with a beefier 12v rail in the case.

It looks like this is a hot issue. Single Rail PSU Solves CrossFire/SLI Problems

I also just posted a news article which approaches this issue from a different angle SLI/Crossfire Power problems. And it's exposing a PSU design issue in that PSUs with a shared Power Plane (5v/12v) dont deliver full power unless both rails in shared powerplane are both loaded.

ExtremeTech : SLI and CrossFire Push Power Supplies to the Limit

What happens is that some power supplies are designed with a shared power plane. According to Tony Ou, of Silverstone technical marketing, in an email:

"I am sure you already know that PC power supplies we have today have three main rails, +3.3V, +5V, and +12V, which are required to power various components. However, what most people don't know is that many power supplies are designed with shared power plane (it is very common to have +5V and +12V rails linked together) to help reduce cost and obtain higher maximum power. If you have seen our retail box for our ST60F 600W model or read the manual, you will see this:

+5V min. load is 10A when +12V output is 30A to 38A
+5V min. load is 15A when +12V output is 38A to 42A

This means that in order for our power supply to generate maximum power for +12V rail optimally, the +5V must also be loaded up. Normally this is not a problem because most systems do draw enough power from both +5V and +12V rails evenly to make cross loading requirement a non-issue. Even when SLI came out in late 2004, the most powerful gaming system at the time would rarely draw more than 30A from the +12V rail..."

In other words, if you balance the loads on the different rails, then you won't have this kind of problem. So we took PCI Express power adapter cable to give this idea a test. These cables consist of two Molex four-pin connectors on one end, and a six-pin PCIe connector on the other end.

Using the same Silverstone 650W power supply as before, we ran the second 512MB 7800 GTX with the adapter connected to a different cable. At this point, everything ran without a hitch. This explains why Falcon Northwest didn't run into this problem. Most people who order a high-end gaming rig don't just ask for SLI or CrossFire. They load them up with multiple hard drives, lots of RAM, and other goodies, which results in a more balanced load.

Sorry for the long winded post....

Omega65

That all being said My DFI Ultra-D is pretty stable with a Opty 165 @ 2.5ghz vCore 1.5, 2GB PC4000, ATI X850XTPE, 4 Hard Drives and dual 16X DVD Burners all being fed by a Silverstone Zeus SST-ST65ZF Quad Rail PSU. The very same one ExtremeTech had the problems with. But I guess I have what the article referred to as "Balanced" system